A Study On The Diagnosis Of N Status Of Cotton Using Image Analysis Methods

This study was focus on the possibility of N nutrition diagnosis of cotton plants by using digital images analysis captured by a commercial digital camera. Field experiment was conducted during 2004 and 2005, on-farm measurements, including LAI, Biomass N absorption was measured to build a statistic model between feature spectrum of images of cotton canopy and biological parameters and then to evaluate the possibility of using spectrum parameters to diagnosis cotton plants N states. Images capture methods was buildup, feature spectrum parameters that N-related was proposed, a good fitness was found between cotton canopy N accumulation and feature spectrum parameters, all of our research showed a sound outlook for the future use. The conclusions as follows:1. The digital image analysis technique can be used as a tool for cotton plant N status diagnosis. The linear correlation was significant between feature spectral parameter which can be obtained from the digital image of cotton's canopy and total N concentration in whole cotton plant; the same was true as the cotton leaves N concentration in different growth period, especially at the day after emergency 98. The feature spectral parameter"g"whose correlation coefficient with N concentration was the highest among other spectrum parameters, between total N concentration in whole plant as well as cotton leaves N concentration in the different growth periods. So g value can be considered as an indicator of N status diagnosis. Also it has a stronger significant correlation between feature spectral parameter and cotton's yield in different growth stage. The significant correlation between the parameter of DGCI and the total N content in whole cotton plant in the whole growth stage which two reflect the same trend is completed. Therefore DGCI value can be used to detect the cotton N status.2. While taken digital image of cotton canopy in the field, the angle between digital camera and ground could be at 60 degree where can get the best regression coefficient between the feature spectral parameter, total N concentration and cotton leaves N concentration. The feature spectral parameters showed no significant difference, when they have captured in the cloudy or sunny day. But it is benefited for analysis digital image in the cloudy day. Any model to take photos work well for the feature spectral parameters showed no significant difference at 4 exposal models. Also the appropriate distance to shoot images is important for the coefficient of variation of the cotton canopy's spectral parameters at 3 .0 meter is smaller than that at 1.0 meter.3. Simulated experiments showed that leaf water concentration influence the feature spectrum parameter greatly. It is proved that the parameter can get maximal scale during the leaf have cutted from the plant body to one hour later. There is a significant correlation between g value and leaf water concentration which also means the feature spectrum...